Comparison of Constant and Non-constant Envelope Signals for Satellite Navigation

The use of constant envelope signals for global navigation satellite systems (GNSS) has a long tradition. These signals allow the satellite payload's high power amplifier (HPA) to be driven in saturation, hence with a very low output power backoff (OBO) and high efficiency. For non-constant envelope signals, the HPA's input power must be carefully chosen near the transition of linear region and saturation region, not to distort the shape of the code-division multiple access (CDMA) waveforms too much - distortions would otherwise lead to a correlation loss at the GNSS receiver. In this paper, we revisit some non-constant envelope CDMA waveforms with raised-cosine (RC) or prolate spheroidal wave function (PSWF) pulse shapes, and try to give a fair comparison with constant envelope waveforms with conventional binary phase-shift keying (BPSK) or frequency-hopping binary offset carrier (FHBOC) spreading. We also address how the operation of the HPA requires a careful trade-off between losses caused by OBO, filtering losses by the payload's output multiplexer (OMUX) and the receiver front-end, as well as receiver correlation loss. Performance is compared in terms of achievable receiver signal-to-noise ratio (SNR) and time of arrival (TOA) estimation error variance. The results suggest that the constant envelope signals' advantage of low OBO is often expended by filtering losses, such that band-limited non-constant envelope spreading waveforms may be an interesting option for future GNSS signal design.